Sales forecasting studies combine inputs gathered from survey research with computer programs to model consumer behavior and predict in-market sales volume. Sales forecasting studies can involve entirely new or established products (e.g., line extensions, flankers, or product re-launches).
Unlike the routine sales projections often made by manufacturing (or sales) for inventory planning purposes, sales forecasting studies are specific pieces of research that are conducted to make critical "go/no go" decisions, such as whether to tool-up a plant. They are usually executed after traditional consumer survey research (e.g., qualitative, concept, product, or positioning studies) has been performed, but prior to a controlled test market or regional/national launch.
The most common stimulus for sales forecasting studies is a concept board, which describes the brand name and finished graphics (for all SKUs), positioning, and pricing. Mechanically, it looks like a board used in a concept test. In addition, because sales forecasting involves comparing product performance to ingoing expectations created by the concept, the final as-marketed product is also needed with finished graphics. Less common are STMs that, in addition to a concept board, use a simulated store with both test and competitive products on-shelf.
Sales Forecasting Designs
Sales forecasts are largely similar to concept-product tests in their basic design. Interviews are conducted in-person, typically in a mall facility. Among those who are neutral-to-positive to the concept, the product is placed for an in-home usage period (usually one week – long enough to allow a respondent to fully experience the product, but not so long as to diminish recall). If there are multiple SKUs to be tested, product is placed based on respondent preference. Respondents are then called back to obtain their reactions to the product, and their likelihood of buying it in the future.
Note also that most sales forecasting systems rely heavily on the use of "norms" – key benchmark questions that have been collected in a standardized way across hundreds of tests. As such, the survey itself cannot be highly customized. Questions must be asked in an exact manner, so that the test can be compared to others in the normative database.
Common measures in both the concept and product phases include:
Purchase intent, or a constant sum measure
Overall rating, open-ended concept likes and dislikes
Value, superiority, uniqueness, believability
Anticipated purchasing frequency and quantity
Anticipated/actual usage situations, HH members who would/did use
Replacement (substitution) vs. addition use
Classification and demographics
Inputs for the sales forecasting model include:
Percent ACV distribution build
Media spending (GRPs), or direct awareness estimates
Percent trial, repeat, and ongoing loyalty
Promotion and trade activity
Purchase cycle, seasonality
Sales forecasting studies are generally conducted among random samples of between 150-300 respondents per cell, with boosts to read diagnostic subgroups of interest as needed. For major brand decisions, greater sampling precision (e.g., samples above 300) may be required for management purposes.
Pros & Cons
Pros: Ability to assess sales potential, with reasonable accuracy, well before significant commitment of company resources; availability of normative benchmarks.
Cons: Expensive and time-consuming; rigid survey format with limited diagnostics; final in-market program often varies from tested concept/product.
Cycle time from field start to results is typically 8-12 weeks. However, stimulus preparation can extend this timeline.
If the idea fails to meet all action standards, Marketing must go back to the drawing board. More likely, the results will be mixed. In this case, additional diagnostic work may – or may not – fix the problem (e.g., additional qualitative, concept, product, or positioning research).
If the results are positive, next steps will depend on the required capital commitment, and an assessment of how positive the results really are. Conservatively, next steps include either a controlled test market (e.g., BehaviorScan), or a "live" in-market test. More aggressively, next steps include a regional roll-out (e.g., a sales zone) or a national launch. In these latter cases, tracking studies should follow to monitor in-market performance.